OAI Overview Michael L. Nelson Old Dominion University Norfolk Virginia, USA mln@cs.odu.edu http://www.cs.odu.edu/~mln/ Bioinformatics Seminar ODU CS 791/891 Feb 3 2003 The Rise and Fall of Distributed Searching • wholesale distributed searching, popular at the time, is attractive in theory but troublesome in practice – Davis & Lagoze, JASIS 51(3), pp. 273-80 – Powell & French, Proc 5th ACM DL, pp. 264-265 • distributed searching of N nodes still viable, but only for small values of N • NCSTRL: N > 100; bad • NTRS/NIX: N<=20; ok (but could be better) The Rise and Fall of Distributed Searching • Other problems of distributed searching (from STARTS) – source-metadata problem • how do you know which nodes to search? – query-language problem • syntax varies and drifts over time between the various nodes – rank-merging problem • how do you meaningfully merge multiple result sets? • Temptations: – centralize all functions • “everything will be done at X” – standardize on a single product • “everyone will use system Y” Universal Preprint Service • A cross-archive DL that that provides services on a collection of metadata harvested from multiple archives – based on NCSTRL+; a modified version of Dienst • support for “clustering” • support for “buckets” • Demonstrated at Santa Fe NM, October 21-22, 1999 – http://ups.cs.odu.edu/ – D-Lib Magazine, 6(2) 2000 (2 articles) • http://www.dlib.org/dlib/february00/02contents.html – UPS was soon renamed the Open Archives Initiative (OAI) http://www.openarchives.org/ Data and Service Providers • Data Providers – publishing into an archive – providing methods for metadata “harvesting” • provide non-technical context for sharing information also • Service Providers – harvest metadata from providers – implement user interface to data • Self-describing archives – Much of the learning about the constituent UPS archives occurred out of band… Even if these are done by the same DL, these are distinct roles Metadata Harvesting • Move away from distributed searching • Extract metadata from various sources • Build services on local copies of metadata – data remains at remote repositories all searching, browsing, etc. performed on the metadata here user individual nodes can still support direct user interaction metadata harvested offline search for “cfd applications” local copy of metadata metadata harvested offline metadata harvested offline metadata harvested offline ... each node independently maintained Result… OAI • http://www.openarchives.org/ • The OAI was the result of the demonstration and discussion during the Santa Fe meeting • Initial focus was on federating collections of scholarly e-print materials… • …however, interest grew and the scope and application of OAI expanded to become a generic bulk metadata transport protocol • Note: – OAI is only about metadata -- not full text! – OAI is neutral with respect to the nature of the metadata or the resources the metadata describes • read: commercial publishers have an interest in OAI too... Santa Fe convention OAI-PMH v.1.0/1.1 OAI-PMH v.2.0 nature experimental experimental stable verbs Dienst OAI-PMH OAI-PMH requests HTTP GET/POST HTTP GET/POST HTTP GET/POST responses XML XML XML transport HTTP HTTP HTTP metadata OAMS unqualified Dublin Core about eprints unqualified Dublin Core document like objects model metadata harvesting metadata harvesting metadata harvesting resources Dublin Core • Dublin Core Metadata Initiative – http://www.dublincore.org/ – from 1994-1995, recognizing the need for simple, interoperable metadata for resource discovery – good overview of metadata & DC: http://www.dlib.org/dlib/january01/lagoze/01lagoze.html – 15 elements (qualifiers possible) Title Creator Subject Description Publisher Contributor Date Typ e Format Identifier Source Language Relation Coverage Rights Overview of OAI Verbs Verb archival metadata harvesting verbs Function Identify description of archive ListMetadataFormats metadata formats supported by archive ListSets sets defined by archive ListIdentifiers OAI unique ids contained in archive ListRecords listing of N records GetRecord listing of a single record most verbs take arguments: dates, sets, ids, metadata formats and resumption token (for flow control) Argument Summary metadataPrefix from until set resumptionToken identifier Identify ListMetadata Formats optional ListSets exclusive ListIdentifiers optional optional optional exclusive ListRecords optional optional optional exclusive GetRecord Error Summary Identify BA ListMetadata Formats BA ListSets BA BRT ListIdentifiers BA BRT CDF NRM NSH ListRecords BA BRT CDF NRM NSH GetRecord BA NMF IDDNE NSH CDF Generate badVerb on any input not matching the 6 defined verbs this is an inversion of the table in section 3.6 of the OAI-PMH specification IDDNE Flow Control • ListSets, ListIdentifiers, ListRecords are all allowed to return partial responses, via a combination of: – resumptionToken – an opaque, archive-defined data string that when passed back to the archive allows the response to begin where it left off • each archive defines their own resumptionToken syntax; it may have visible semantics or not – 503 http status code – “retry after” • up to the harvester to understand this code and respect it, and up to the archive to enforce it resumptionToken scenario: harvesting 277 records in 3 separate 100 record “chunks” ListRecords harvester Records 1-100, resumptionToken=AXad31 ListRecords, resumptionToken=AXad31 Records 101-200, resumptionToken=pQ22-x ListRecords, resumptionToken=pQ22-x Records 201-277 RDBMS OAI Links & Demos • Data providers – not really meant for end-user interaction, but Suleman’s “Repository Explorer” is an excellent tool • http://purl.org/net/oai_explorer • ~100 registered data providers – http://oaisrv.nsdl.cornell.edu/Register/BrowseSites.pl – many being used for internal purposes; not registered • Service providers – Arc, the first known SP harvesting from OAI data providers • http://arc.cs.odu.edu/ • ~20 registered service providers – http://www.openarchives.org/service_provider/oai_sp.htm – several more known to be in testing or creation Field of Dreams • It should be easy to be a data provider, even if it makes more work for the service provider. – if enough data providers exist, the service providers will come (DPs >> SPs) • Open-source / freely available tools – “drop-in” data providers: • industrial strength: http://www.eprints.org/ • personal size: http://kepler.cs.odu.edu/ – tools to make your existing DL a data provider: • http://www.openarchives.org/tools/tools.htm • also: OAI-implementers mailing list / mail archive! – service providers: • only bits and pieces currently publicly available... OAI Observation: Front-End Only • No input/registry mechanism – OAI harvesting protocol is always a front-end for something else • filesystem, Dienst, RDBMS, LDAP, etc. – convenient for pre-existing DLs, but does not address “new” DLs • e.g., “we want to do OAI” • Bounds the scope of OAI – responsibilities and domain of OAI are still be discussed – tension between functionality and simplicity OAI Observation: No T&C • No terms & conditions provisions in protocol – assumes all metadata has uniform access rights • how to restrict metadata to certain hosts? – introducing T&C would increase the scope of application, but at the expense of simplicity • how expensive do we want to make a “just-a-front-end protocol” ? • maybe T&C is a good application for sets? OAI Observation: No T&C • Possible to use multiple OAI servers in a DMZ-like configuration… OAI requests from arbitrary hosts Public OAI Server OAI requests from trusted hosts Private OAI Server Source database could even use a separate copy of the database… OAI Observation: No T&C • Possible to use OAI harvesting protocol in closed, restricted systems OAI 1 OAI 2 OAI 4 OAI 3 all OAI requests originate from these 4 DLs OAI Observation: Monolithic • An OAI server has no protocol-defined concept of “other” OAI servers – backups, mirrors, etc. have to be resolved outside of the scope of OAI • scope vs. complexity again – fully connected graph of DLs harvesting from each other is unnecessary • cf. web crawlers vs. “gathers” in U of Colorado’s Harvest System – 3rd party harvesting interfaces raise more T&C and data coherency issues 302 Load Balancing • Interactive users on main DL machine should not be impacted by metadata harvesting – don’t take deliveries through the front door – not part of the protocol; defined outside the protocol if load > 0.05 redirect request http://blah/oai/?verb=ListIdentifiers harvester HTTP Status Code 302 OAI Server naca.larc.nasa.gov/oai/ http://blah/oai/?verb=ListIdentifiers <?xml version=“1.0” encoding=“UTF-8”?> … <ListIdentifiers> … </ListIdentifiers> OAI Server buckets.dsi.internet2.edu/naca/oai/ OAI Observation: Data Coherency • In the interest of OAI implementer simplicity, several issues are left for the service provider to interpret – what is an update vs. addition? • in the NACA OAI interface, they are reported as the same and its up to the harvesting system to figure it out – deletions? • it is currently optional for OAI systems to mark records as deleted or not… – still left to the harvester to interpret OAI Observation: Harvest Model • Frequency of harvests – all-at-once harvests? • initial harvest • resolving data coherency – frequent incremental harvests? • far more efficient for both service and data providers • Webcrawling vs. digital library models – webcrawlers: little to no a priori information about target – DLs: frequent harvesting of a small number of known targets • Realization: we know very little about how harvesting behavior… – are we optimizing for all-at-once, when incremental will be more common? Other Uses For the OAI-PMH • Assumptions: – Traditional DLs / SPs will continue on their present path of increasing sophistication • citation indexing, search results viz, personalization, recommendations, subject-based filtering, etc. – growth rates remain the same (5x DPs as SPs) • Premise: OAI-PMH is applicable to any scenario that needs to update / synchronize distributed state – Future opportunities are possible by creatively interpreting the OAI-PMH data model OAI-PMH Data Model set-membership is item-level property item = identifier Dublin Core metadata resource all available metadata about David MARC metadata SPECTRUM metadata item records record = identifier + metadata format + datestamp Typical Values • repository • resource – scholarly publication item – all metadata (DC + MARC) record – a single metadata format datestamp – last update / addition of a record metadata format – bibliographic metadata format set – originating institution or subject categories • • • • • – collection of publications Repositories… • Stretching the idea of a repository a bit: – contextually sensitive repositories • “personalization for harvesters” • communication between strangers, or communication between friends? – OAI-PMH for individual complex objects? • OAI-PMH without MySQL?! – Fedora, Multi-valent documents, buckets – tar, jar, zip, etc. files Resource • What if resource were: – computer system status • uptime, who, w, df, ps, etc. – or generalized “system” status • e.g., sports league standings – people • personnel databases • authority files for authors Item • What if item were: – software • union of versions + formats – all forms of metadata • administrative + structural • citations, annotations, reviews, etc. – data • e.g., newsfeeds and other XML expressible content – metadataPrefixes or sets could be defined to be different versions Record • What if record were: – specific software instantiations / updates – access / retrieval logs for DLs (or computer systems) – push / pull model inversion • put a harvester on the client behind a firewall, the client contacts a DP and receives “instructions” on how to submit the desired document (e.g., send email to a specified address) Datestamp • semantics of datestamp are strongly influenced by the choice of resource / item / record / metadataPrefix, but it could be used to: – signify change of set membership (e.g., workflow: item moves from “submitted” to “approved”) – change datestamp to reflect access to the DP • e.g., in conjunction with metadataPrefixes of “accessed” or “mirrored” metadataPrefix • what if metadataPrefix were: – instructions for extracting / archiving / scraping the resource • verb=ListRecords&metadataPrefix=extract_TIFFs – code fragments to run locally • (harvested from a trusted source!) – XSLT for other metadataPrefixes • branding container is at the repository-level, this could be record- or item-level Set • sets are already used for tunneling OAI-PMH extensions (see Suleman & Fox, D-Lib 7(12)) • other uses: – in aggregators, automatically create 1 set per baseURL – have “hidden” sets (or metadataPrefix) that have administrative or community-specific values (or triggers) • set=accessed>1000&from=2001-01-01 • set=harvestMeWithTheseARGS&until=2002-0505&metadataPrefix=oai_marc Interesting Services • DP9 – gateway to expose repository contents in HTML suitable for web crawlers • Celestial – OAI “cache”, also 1.1 -> 2.0 converter • Static (mini-) repositories – XML files, based on OLAC work • OpenURL metadata format registries – record = metadata format